The De‐biasing Effect of Incidental Anger on Other‐provided Anchors

The current research investigates the effect of incidental anger on anchoring bias. We hypothesized that feeling angry will make people less influenced by other‐provided anchors because of the moving against action tendency associated with anger. That is, individuals in an angry state will be likely to perceive a given anchor as a viable target for their desire to attack and actively seek out anchor‐inconsistent information, thereby committing less anchoring bias. To examine our hypothesis, in Study 1, we manipulated emotions using film clips and administered a general knowledge task with other‐provided anchors. As predicted, participants in the anger condition showed less anchoring bias to the other‐provided anchors than those in the sad or neutral condition. Study 2 replicated the finding with a different emotion manipulation technique and different anchoring questions. More important, consistent with the moving against action tendency explanation, we also found that people in an angry state committed more anchoring bias for self‐generated anchors, compared with people in a sad or neutral state. Copyright © 2011 John Wiley & Sons, Ltd. key words discrete emotions; anger; anchoring bias; anchor source; action tendency People tend to search for a reference point when information is insufficient or ambiguous. For example, when buying a house, people usually check the prices of other houses in the same neighborhood to get a sense of whether the list price is appropriate (Northcraft & Neale, 1987). This strategy seems reasonable as long as the reference point offers valuable, relevant information. Yet, even random, irrelevant numbers can influence judgments (Tversky & Kahneman, 1974). This phenomenon — “anchoring heuristic” — has been defined as “the assimilation of a numeric estimate to a previously considered standard” (Mussweiler, Strack, & Pfeffer, 2000). Anchoring is a robust phenomenon. It has been reported in numerous contexts including general knowledge questions (Strack & Mussweiler, 1997; Tversky & Kahneman, 1974), evaluation of gambles (Chapman & Johnson, 1999), probability estimates (Plous, 1989; Wright & Anderson, 1989), trait inference (Gilbert, 1989; Kruger, 1999), and egocentric biases (Gilovich, Medvec, & Savitsky, 2000). Also, the effects remain strong when the anchor value is extreme (Strack & Mussweiler, 1997), or even when people know the anchor value was randomly selected by either a wheel of fortune (Tversky & Kahneman, 1974) or generated from their social security number (Wilson, Houston, Etling, & Brekke, 1996), both of which are irrelevant to the estimate. More surprising, anchoring effects do not disappear when participants are motivated to improve their accuracy (Wilson et al. 1996) or when they have expertise in the field (Wright & Anderson, 1989). To date, only a few studies have discovered factors that can alleviate anchoring effects, and those that have done so are primarily based on cognitive strategies such as reducing overconfidence in one’s belief by intentionally considering belief‐inconsistent evidence (Griffin, Dunning, & Ross, 1990; Koriat, Lichtenstein, & Fischoff, 1980; Lord, Lepper, & Preston, 1984). Some prior research has investigated *Correspondence to: Heajung Jung, University of California, Los Angeles, CA, USA. E‐mail: [email protected] Copyright © 2011 John Wiley & Sons, Ltd. whether anchoring can be affected by incidental emotions (Bodenhausen, 1993) — emotions that are caused by a prior event and are unrelated to the current situation. These findings show that incidental sadness is associated with more anchoring (Bodenhausen, Gabriel, & Lineberger, 2000; Englich & Soder, 2009). However, prior research has not examined whether anger, a commonly experienced emotion (Averill, 1982), might also affect anchoring. Anger is an interesting emotion to investigate in that it is also negative in valence but sharply contrasts with sadness in its characteristics. Anger is defined as a negative affective state “arising from the blockage of movement toward a desired goal” (Carver & Harmon‐Jones, 2009). Specifically, anger is often directed at another person who engages in intentional acts to disturb one’s efforts to achieve a goal and is thus perceived to be blameworthy for negative consequences (Frijda, 1988; Ortony, Clore, & Collins, 1988). Anger is associated with a behavioral readiness — or action tendency — to approach and act against (Carver & Harmon‐Jones, 2009; Frijda, Kuipers, & ter Schure, 1989), compared with sadness that is characterized by avoidance and inaction. We expect that anger will have the opposite effect on anchoring bias to sadness because of this moving against action tendency. Selective accessibility model and the consider‐the‐ opposite strategy In a traditional, two‐step anchoring task (Tversky & Kahneman, 1974), people are first asked to judge whether their own estimate is greater or less than an anchor value (e. g., Is the Mississippi River longer or shorter than 2000 miles?) and then to give their final estimate (e.g., What is the length of the Mississippi River?). According to the selective accessibility model (Mussweiler & Strack, 1999), anchoring is driven by selective increases in the accessibility of anchor‐ consistent knowledge about the target. More specifically, when people are asked to compare a target object with a 436 Journal of Behavioral Decision Making given anchor, they engage in a hypothesis‐testing procedure where they try to selectively retrieve from their memory knowledge supporting the hypothesis that their estimate equals the anchor (selectivity). Later when they are asked to give a final estimate, the hypothesis‐consistent knowledge retrieved earlier is easily activated, therefore influencing the final judgment (accessibility). If anchoring is caused by retrieving anchor‐consistent evidence from memory, encouraging people to search for anchor‐inconsistent evidence should make them less influenced by the given anchors (Mussweiler et al., 2000). This strategy to “consider‐the‐opposite” asks participants to think about reasons why anchor values might be inappropriate or to list three arguments against the anchor, and these manipulations significantly weakened the anchoring bias. We believe anger will serve as a catalyst triggering people to argue against the anchor via its inherent moving against action tendency. Anger and anchoring bias From an evolutionary and functionalistic perspective, discrete emotions are characterized by unique action tendencies (Frijda, 1988; Frijda et al., 1989). The action tendencies are believed to have developed as strategies for dealing with threats in human life and have become internalized as automatic responses. They can be broadly categorized into an approach tendency (i. e., seek out interaction with a target) and an avoidance tendency (i.e., withdraw from a target) (Carver & White, 1994). Anger belongs to the former category (Carver & Harmon‐ Jones, 2009), and the approach tendency of anger entails hostile goals. That is, anger is associatedwith a tendency to go against or oppose a target (Frijda, 1988), compared with other negative emotions’ action tendencies such as moving away (fear) or doing nothing but crying (sadness). The action tendencies triggered by certain emotions can linger, thereby leading individuals to respond to a subsequent situation in a manner consistent with the activated action tendencies (Young, Tiedens, Jung, & Tsai, 2011). The moving against action tendency of anger has been found to lead to an active search for a target against which aggression can be directed. Prior research has shown that angry individuals tend to pay more attention to hostile stimuli than neutral stimuli (Cohen, Eckhardt, & Schagat, 1998; Smith & Waterman, 2003). Young et al. (2011) demonstrated that when angry, individuals tend to seek out attitude‐inconsistent information out of their desire to argue against it. Our question is whether an anchor can serve as a viable target. According to the selective accessibility model, individuals test the validity of an anchor value provided by someone else. It is possible that angry individuals will perceive the anchor as a target for their desire to attack and actively seek out information refuting the validity of the given anchor, consequently committing less anchoring bias. In this paper, we will test whether anger can reduce anchoring bias than sad and neutral emotional states. Sadness will serve as a good comparison state because it differs from anger in its action tendency, and its effect on anchoring has been already examined. We hypothesize that when an anchor value is provided by someone else, people Copyright © 2011 John Wiley & Sons, Ltd. in an angry state will be less susceptible to anchoring than people feeling sad or neutral.